The effects of water storage changes in China's Three Gorges Reservoir on in situ gravity measurements

The gravity field at locations near China's Three Gorges Reservoir (TGR) changes primarily because of variations in the reservoir's water level. Those variations also give rise to surface vertical displacements, level plane changes, fluctuations in groundwater, and secondary geological hazards. Since 2008 there have been four consecutive years of experimental impoundment in TGR, with the lake surface reaching an elevation of 175 m at full capacity. The reservoir's water level varies seasonally between 145 m and 175 m from summer to winter. When the water level is near its 175 m maximum, there are also irregular water level variations of the order of 1 meter, that occur at characteristic periods of a few weeks. Here, we describe the use of in situ measurements from gravimeters near the TGR, to detect gravity field variations caused by these various changes in TGR water impoundment. The gravitational effects caused by a lake level change can be separated into the direct gravitational attraction of the water, and secondary effects caused by the Earth's deformation in response to the change in surface loading. We model the direct attraction by dividing the lake into a large number of thin, vertical columns of water, and finding the gravitational acceleration caused by each column. We model the contributions related to the Earth's deformation by using elastic loading Green functions as described by Farrell (1972). The lake level variations are determined from tide gauge measurements at each end of the TGR. The modelled results show that the gravity along the edge of the lake can vary seasonally by up to a few hundred micro-gals if the gravimeter is situated well above the lake surface. The predicted variations caused by the ~1 meter fluctuations in lake level can be on the order of a few to 10 micro-gals. We use data from two years of repeat absolute gravity (A10) campaigns at several locations, and from roughly 5 months of continuous relative gravity (gPhone) measurements at a single site near the lake, to compare with the model predictions. For the seasonal variation, the agreement between the modelled and observed amplitudes is good, but there are not yet enough data to assess the agreement in phase. For the ~1 meter variability that can occur near maximum lake level, the observed signal has a somewhat larger amplitude than the predicted signal, and shows a consistent phase lag of a few days. We propose that these latter differences could indicate variability in the underlying groundwater beneath the instrument.